Customer Analytics: A Guide to Getting Started (part 6)
October 29th, 2008 by Doug Bright(Read the entire Customer Analytics: A Guide to Getting Started series)
The situation so far: We have a glut of women’s sweater inventory that we need to unload. We decided to identify the customers who are most likely interested in purchasing women’s sweaters so that we can send full priced offers to them while offering deep discounts to the rest of the customer base.
In part 4 we produced the relevant data and had our data modeler (or software solution) produce a list of customers and their relative probabilities of responding to a sweater offer.
Now it’s time to execute our plan. We first partition the customers into two groups: likely to buy at full price and unlikely to buy at full price. ”Likely” is a relative term here — we don’t mean that a “likely” buyer has a probability of buying greater than 50%. In fact, the likely group may have only a 1% response rate. But if the unlikely group has only a 0.1% response rate, it means the likely group is 1000% more likely to respond than the unlikely group. It’s up to us to make an initial judgment call on what cutoff probability to use for likely vs. unlikely.
A few days before we send the offers (let us assume we’re sending the offers by email), we want to run a test to see if we can improve upon our intuition for what is a good cutoff probability. We do this by pulling out two samples of customers from each group (four total) for the test. The size of the group depends on the expected response rate and how much certainty you need to feel comfortable acting on the results (we discussed what is an appropriate test group size in an earlier post about holdout testing. You can also use the excellent sample size calculator from Raosoft). We then send each sample group an offer, either with a discount or without according to the following 2×2:
| Discount | No Discount | |
| Likely Buyers | ||
| Unlikely Buyers |
Within a few days of sending the test offers we fill in the boxes with the click-through rates of each group. Using this information we can decide if we want to readjust the cutoff percentage of the groups. That’s what we’ll do next time.
Tags: Customer Analytics, getting started
Add Personalized Product Recommendations to Your eCommerce Site
Istobe is a powerful recommendation engine that makes it easy to add recommendations to your eCommerce site. How powerful? Shoppers who click on Istobe recommendations spend 20-50% more than the average visitor.
